In a recent article in this journal, Roberts suggests a semantic method for distinguishing ambiguity and generality, a method which is intended to avoid the problems that others such as Zwicky and Sadock, Hintikka, and McCawley have found in making such a decision. Roberts claims that his test derives its validity from the observation that an ambiguous expression has a disjunction of meanings, whereas a general expression has but one meaning.

In a recent article in this journal, Roberts suggests a semantic method for distinguishing ambiguity and generality, a method which is intended to avoid the problems that others such as Zwicky and Sadock, Hintikka, and McCawley have found in making such a decision. Roberts claims that his test derives its validity from the observation that an ambiguous expression has a disjunction of meanings, whereas a general expression has but one meaning.

A Distinction between "internal" and "external" evidence in linguistics is illustrated, and two occasions on which the distinction arises are identified: in the division of labor between linguistics and other fields, and in the choice among alternative descriptions. Assumptions which would bias generative linguists both away from and towards external evidence are explored. Examples from phonological and syntactic analyses are contrasted, and speculations are made as to why evidence should be differently used in phonology and syntax. A prescription favoring external (...) evidence is tempered by the need for credible assumptions linking linguistics to other fields. (shrink)